b10301

ggml-org/llama.cppb10301Aug 7, 2026by github-actions[bot]same day as b10299

What changed since b10299

low risk

Maintenance delta: CUDA unused-variable/function warning cleanup plus additional prebuilt backend binaries.

New features

  • New Windows x64 prebuilt binaries for CUDA 12.4 and CUDA 13.3.
  • New Windows x64 prebuilt binaries built against OpenVINO 2026.2.1.
  • New Windows x64 SYCL binaries in both FP32 and FP16 variants.
  • New openEuler packages targeting 310p and 910b with ACL Graph support.

Improvements

  • CUDA sources cleaned up so unused variables and functions no longer emit compiler warnings.
  • Broader artifact matrix means fewer users need to build from source for their backend/arch combination.

Bug fixes

  • Resolved noisy CUDA build warnings that could obscure real diagnostics in CI and local builds.

Upgrade notes

This is a two-build maintenance step with no API, CLI, or model-format changes, so a straight swap of binaries or a rebuild is sufficient. If you consume prebuilt artifacts, verify the new CUDA 12.4/13.3, OpenVINO, SYCL, and openEuler variants match your driver and toolkit versions. The Metal threadgroup fix from b10299 remains included.

Analyzed Oct 30, 2026

Upgrading?

Release notes summary

A maintenance release focusing on fixing CUDA compiler warnings regarding unused variables and functions. It includes standard binaries for macOS, Linux, Android, Windows, and openEuler across various architectures and hardware backends.

Key Highlights

  • Fixed CUDA warnings for unused variables and functions (#26688)
  • Added binaries for Windows x64 (CUDA 12.4 and 13.3)
  • Added binaries for Windows x64 (OpenVINO 2026.2.1)
  • Added binaries for Windows x64 (SYCL FP32 and FP16)
  • Added binaries for openEuler (310p and 910b with ACL Graph)

Release Notes (excerpt)

<details open> cuda: fix warnings for unused variable/function (#26688) </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - Ubuntu x64 (CPU) - Ubuntu arm64 (CPU) - Ubuntu s390x (CPU) - Ubuntu x64 (Vulkan) - Ubuntu arm64 (Vulkan) - Ubuntu x64 (ROCm 7.2) - Ubuntu x64 (OpenVINO) - Ubuntu x64 (SYCL FP32) - Ubuntu x64 (SYCL FP16) **Android:**…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories